Two novel solid-state instrumental designs are described for use in the noninvasive measurement of blood glucose. Both instruments implement transmission measurements of near-infrared light through a tissue sample and allow the encoding of multivariate calibration models in hardware. Each instrument is based on the sequential measurement of pre-defined combinations of near- infrared wavelengths that are specified in the form of filter bandpass functions. In the two designs, these filter profiles are generated with either a digital micromirror array (MMA) or an acousto-optic tunable filter (AOT F). The MMA allows the specification of filter profiles of arbitrary complexity, while the AOTF allows the generation of filters with simpler square-wave shapes. A computational model and numerical optimization procedures are described for use in defining optimal filter profiles that result in the measurement of unique spectral patterns that are selective for blood glucose. These filter profiles are analogous to the components extracted by multivariate calibration models from contiguous spectral data acquired by a conventional spectrometer. Performing these discrete filter-based measurements gains both throughput and multiplex advantages, and allows the creation of rugged instrumental designs that are compatible with noninvasive blood glucose measurements made by diabetic patients outside the controlled environment of the laboratory. Both in vitro experiments and in vivo studies with a rat model are proposed for use in characterizing the performance and analytical utility of these instrumental designs. The experimental and computational work will be directed to the overall goal of developing a noninvasive nocturnal hypoglycemic alarm.